If Intercom's Fin only answers knowledge base questions for you, check one thing before switching: Fin (now part of Salesforce) can resolve multi-step tickets through Procedures, data connectors and MCP, but only once you build those and turn on user authentication. If you have done that and still need more, the AI agents that resolve complex tickets inside or alongside Intercom are Lorikeet and Pluno, while Decagon, Sierra, Ada and Salesforce Agentforce are larger platform changes. The right pick depends on whether your hard tickets need backend actions, strict identity checks, voice, or a full audit trail.

Every vendor fact below was checked on the vendor's own public pages on 30 October 2026 and is linked to its source.

Key takeaways

• Fin is not FAQ-only by design. Intercom's docs say Procedures let Fin resolve "account troubleshooting, damaged order claims, or identity verification" end to end, with if/else logic, code and system access.

• The usual blocker is setup, not the model. Fin's data connectors cannot go live for users until JWT user authentication is configured, so many workspaces stay on knowledge answers.

• Complex tickets need four things: actions in backend systems, authentication, policy logic and an audit trail. Test any agent on all four with your own escalated tickets.

• Billing rules differ. Fin charges $0.99 per outcome, including a Procedure that ends in a configured handoff. Lorikeet and Pluno do not charge for escalations.

Which AI agents can resolve complex tickets in Intercom?

Lorikeet connects to Intercom through an admin-approved authorization flow and handles conversations across Intercom channels with access to customer profiles and history, handing complex cases to your team with full context (Intercom integration page). Other options:

• Fin Procedures combine natural-language steps with conditions, code and data connectors or MCP (Intercom help).

• Pluno describes itself as an AI support agent for complex B2B tickets that resolves tickets in your help desk (Pluno pricing).

• Decagon runs complex workflows as natural-language Agent Operating Procedures (Decagon AOPs).

Can Fin resolve complex tickets, or only answer from the knowledge base?

Fin can resolve complex tickets when it is configured to. Intercom's help centre says Procedures "enable Fin to resolve complex queries like damaged order claims or account troubleshooting from start to finish", combining natural-language instructions with deterministic controls (Fin Procedures explained). You can add if/else conditions, use code "for precision" in eligibility checks or record updates, reuse sub-procedures, connect tools such as Stripe or Shopify through data connectors or MCP, and run simulations that judge whole conversations against your success criteria. Fin's product page adds handoff controls and full regression testing on saved simulations (fin.ai/procedures).

There are documented limits. Procedure actions run one at a time ("No parallel processing"), Procedures cannot be called directly from a Workflow step, and simulations work only inside Procedures. None of these stops most complex tickets, but they matter for designs that query several systems at once.

Why is Fin only answering FAQs for you?

In most workspaces the cause is one of three gaps:

1. No data connectors. Each data connector is an API call Fin can choose to make; "you can connect any system which has an API" (Fin data connectors). Without them, Fin only has your content.

2. No user authentication. The same article says "user authentication must be configured in order to set a data connector live for users". Fin's JWT guide warns that on an unsecured Messenger someone could "spoof the identity of another user" (Fin JWT guide).

3. No Procedures. Workflows in Intercom are visual reply-button flows; Procedures are the AI-driven automations meant for multi-step processes.

If these are missing, building them is usually the cheapest next step. If they are in place and complex tickets still escalate, the problem is fit, and it is time to compare agents.

What does an AI agent need to resolve complex tickets?

Complex tickets are the ones where the answer depends on the customer's own data and a rule. The CFPB found chatbots "may be useful for resolving basic inquiries, but their effectiveness wanes as problems become more complex" (CFPB), and Gartner found only 14% of customer service issues are fully resolved in self-service (Gartner). Any agent you shortlist should show:

• Actions in backend systems: refunds, plan changes, address updates, chained in one conversation.

• Authentication: NIST SP 800-63B's AAL2, for example, requires "proof of the possession and control of two distinct authentication factors" (NIST SP 800-63B). Decide which tickets need that level before the agent touches account data. See our guide to verifying identity when an AI agent contacts support.

• Policy logic: eligibility, dates and jurisdiction rules enforced in code, not left to the model.

• Audit and testing: every tool call and decision replayable, and changes tested before release. OWASP notes it is "unclear if there are fool-proof methods of prevention for prompt injection", so limit what each action can reach (OWASP LLM01).

How we chose and checked these

We included agents that publish a way to take actions in other systems, and we flagged whether each one works inside Intercom or replaces it. Every vendor fact was checked on the vendor's own public pages on 30 October 2026. Where a vendor does not publish an Intercom integration or a price, we say "Not published".

Lorikeet publishes this guide and is one of the tools listed. We have tried to describe every product, including our own, using only what each vendor publishes, and we link to the source for each claim.

AI agents for complex tickets compared

1. Fin with Procedures: fix it in place first

If you are on Intercom, the lowest-effort route to complex resolution is to build Procedures, data connectors and JWT authentication in Fin. Fin is now part of Salesforce, which completed the acquisition on 10 September 2026 and says Fin resolves queries across live chat, email, WhatsApp, SMS, voice and Slack (Salesforce announcement). Pricing is $0.99 per outcome with no setup, integration or platform fees (Fin pricing).

Best for: Intercom teams whose complex tickets need a handful of API lookups and clear rules.

Trade-offs: a billable outcome includes a Procedure "configured to end in a handoff to a human or a workflow", so heavy escalation designs still cost money. Sequential actions and Procedure-only simulations are constraints to design around.

2. Lorikeet: complex and regulated tickets inside Intercom

This agent plugs into Intercom and resolves tickets that need work in other systems. Workflows act through APIs, webhooks and MCP servers; payments and identity checks run as deterministic code the agent can call but never alter; and you replay historical tickets before launch (how it works). You can start in a conservative mode and expand automation as confidence grows. Channels include chat, email, voice, SMS and WhatsApp.

Linktree plugged the agent into Intercom to work alongside its human agents, with early workflows for account access investigations and refund and cancellation flows tied to 72-hour policies, and cut first response time to 1 minute. Its story says it had also looked at Fin and Decagon.

Pricing: Start $2,100 a month and Scale $5,100 a month, paid annually, plus $0.99 or $0.90 per resolved chat or email ticket; escalations are not charged; 30-day free trial (pricing).

Limits: the plan fee applies before any resolutions, which makes it expensive for small or mostly informational queues, and it does not hold PCI DSS, so card capture stays in your own payment flow (trust).

3. Pluno: complex B2B and technical tickets in Intercom

Pluno positions itself as "the AI support agent for complex B2B tickets" and learns from past resolved tickets and connected tools (pluno.ai). On its own Intercom guide, Pluno says it integrates directly with Intercom, searches past conversations, Slack, Jira and Confluence, takes actions through custom APIs, and simulates on past Intercom conversations before go-live (Pluno guide). Pricing starts at €249 a month plus €0.90 per resolution; escalations are never counted; 14-day free trial with no credit card (Pluno pricing).

Trade-offs: Pluno itself lists no voice channel, and its focus is B2B and technical products rather than regulated consumer workflows.

4. Decagon: replace the agent layer at enterprise scale

Decagon defines workflows as Agent Operating Procedures that combine natural language "with the reliability of code", with Git-based versioning for technical teams (Decagon AOPs). It covers chat, voice and email (decagon.ai). We could not find a published Intercom integration page, so confirm how it connects to your inbox. Pricing is not published.

5. Sierra: enterprise agent with payment compliance

Sierra connects to systems of record to complete tasks "like processing an insurance claim, returning an order, or originating a mortgage", and works across voice, chat, email and WhatsApp (Sierra product). Agent Studio offers 40+ pre-built integrations and real-time traces of every decision (Agent Studio). It prices on outcomes without a public rate card, and an Intercom connector is not listed on the pages we checked.

6. Ada: SOP automation across voice, messaging and email

Ada offers Playbooks to "automate your complex SOPs with agentic AI" and Ada Computer to give the agent tools to complete customer tasks (ada.cx). It suits large enterprises; pricing and an Intercom integration page are not published on its homepage.

7. Salesforce Agentforce: if you are moving to Salesforce

Agentforce is relevant only if your service data lives in Salesforce. It publishes $2 per conversation, Flex Credits or per-user licensing (Agentforce pricing).

How to choose an AI agent that can resolve complex tickets

1. Pull 200 recent escalated Fin conversations and tag why each escalated: missing data, missing action, identity check, policy judgement or customer request for a person.

2. If most are missing data or actions, build Fin data connectors and Procedures first.

3. If most need strict identity checks, regulated policy logic or voice, test Lorikeet on the same set. For B2B technical troubleshooting, test Pluno.

4. Ask every vendor to replay the set and show each tool call on the failures, and to put its billing definition of a resolution in writing.

For a wider comparison, see the best AI agents inside Intercom and Fin vs Lorikeet.

What still needs a human?

Keep people on judgement calls: hardship, complaints with regulatory weight, fraud and disputes that need investigation, and anything clinical. The NIST AI Risk Management Framework is a useful reference for deciding which decisions need human oversight. Escalate with context when the customer asks for a person, when the agent stops making progress, or when policy requires a human decision, and keep compliance sign-off on new workflows with your compliance team.

If your Intercom queue fits the complex profile above, start a 30-day Lorikeet free trial and test it on your own escalated tickets.